I'm going to assume that $a$ and $b$ are non-zero real numbers (for simplicity).
In any proof, we always begin by checking definitions. In this case, the definition of $x^{-1}$ is: the unique value $y$ such that $x \cdot y = 1$.
So we know that $(ab) \cdot (ab)^{-1} = 1$. We also know that
$\begin{equation}
\begin{split}
(ab) \cdot (a^{-1} \cdot b^{-1})
&= ((a \cdot b) \cdot a^{-1}) \cdot b^{-1} \\
&= ((b \cdot a) \cdot a^{-1}) \cdot b^{-1} \\
&= (b \cdot (a \cdot a^{-1})) \cdot b^{-1} \\
&= (b \cdot 1) \cdot b^{-1} \\
&= b \cdot b^{-1} \\
&= 1
\end{split}
\end{equation}
$
So since $(ab)^{-1}$ is the only $y$ such that $(ab) \cdot y = 1$, and it is also true that $(ab) \cdot (a^{-1} b^{-1}) = 1$, it must be the case that $(ab)^{-1} = a^{-1} b^{-1}$.
Notice that we always start with the definition. Once you know the definition of a thing, you can begin to derive its properties.